Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
Gemini 3D Traffic Twins
Reconstruct messy traffic camera feeds into interactive 3D twins, infer flows, and generate "what-if" videos showing safer city designs.
Toronto traffic feels chaotic but our cameras already see everything.
In this demo, I’ll show how we can take a raw traffic camera frame or short clip from a messy intersection, and use Gemini 3 Pro Vision, SAM 3D, and Veo 3.1 to:
Reconstruct the scene as an interactive 3D traffic “digital twin”
Infer flows, queues, and conflict zones from just video
Generate cinematic “what-if” videos of safer, people-first designs
It’s a very opinionated look at how multimodal models + generative video can go beyond chatbots and actually help cities reason about congestion, safety, and design all starting from the same grainy camera feeds we already have.